mirror/maigret
1
0
Fork 0
mirror of https://github.com/soxoj/maigret.git synced 2026-05-07 06:24:35 +00:00
Code Issues Packages Projects Releases Wiki Activity

Merge pull request #142 from soxoj/photo-sites-1

Browse Source 
Photo sites added
This commit is contained in:
soxoj
2021-05-09 18:24:33 +03:00
committed by GitHub
parent b7bdd71cf0 acc751ff98
commit ead048af93
2 changed files with 111 additions and 11 deletions
Download Patch File Download Diff File Expand all files Collapse all files
maigret/resources/data.json
+107 -11
View File
@@ -1253,8 +1253,8 @@
},
"Artstation": {
"tags": [
"ca",
"us"
"art",
"stock"
],
"checkType": "status_code",
"alexaRank": 1444,
@@ -3120,7 +3120,8 @@
},
"CreativeMarket": {
"tags": [
"us"
"art",
"stock"
],
"errors": {
"/cdn-cgi/scripts/hcaptcha.challenge.js": "Captcha detected"
@@ -3804,7 +3805,6 @@
"tags": [
"photo",
"art",
"us",
"stock"
],
"checkType": "status_code",
@@ -5159,8 +5159,7 @@
},
"Fotothing": {
"tags": [
"gb",
"in"
"photo"
],
"checkType": "message",
"absenceStrs": [
@@ -6915,7 +6914,7 @@
"ImageShack": {
"tags": [
"photo",
"in"
"sharing"
],
"checkType": "response_url",
"alexaRank": 9830,
@@ -7887,7 +7886,8 @@
},
"Lightstalking": {
"tags": [
"in"
"photo",
"forum"
],
"checkType": "status_code",
"alexaRank": 527548,
@@ -24213,7 +24213,7 @@
},
"fancy.com": {
"tags": [
"in"
"shopping"
],
"engine": "engine404",
"alexaRank": 31681,
@@ -25289,11 +25289,14 @@
"absenceStrs": [
"<html><head><title>404 \u0421\u0442\u0440\u0430\u043d\u0438\u0446\u0430 \u043d\u0435 \u043d\u0430\u0439\u0434\u0435\u043d\u0430</title></head><body><h2>404 \u0421\u0442\u0440\u0430\u043d\u0438\u0446\u0430 \u043d\u0435 \u043d\u0430\u0439\u0434\u0435\u043d\u0430</h2></body></html>"
],
"alexaRank": 5270222,
"alexaRank": 5278356,
"urlMain": "https://wblitz.net",
"url": "https://wblitz.net/stat/ru/{username}",
"usernameClaimed": "lucklev12",
"usernameUnclaimed": "noonewouldeverusethis7"
"usernameUnclaimed": "noonewouldeverusethis7",
"tags": [
"gaming"
]
},
"unc.ua": {
"tags": [
@@ -27504,6 +27507,99 @@
"tags": [
"photo"
]
},
"iStock": {
"absenceStrs": [
"subheading"
],
"presenseStrs": [
"collectionName"
],
"url": "https://www.istockphoto.com/ru/portfolio/{username}",
"urlMain": "https://www.istockphoto.com",
"usernameClaimed": "leowilde",
"usernameUnclaimed": "noonewouldeverusethis7",
"checkType": "message",
"alexaRank": 461,
"tags": [
"photo",
" stock"
]
},
"Brusheezy": {
"absenceStrs": [
"masthead"
],
"presenseStrs": [
"username",
" user-name"
],
"url": "https://www.brusheezy.com/members/{username}",
"urlMain": "https://www.brusheezy.com",
"usernameClaimed": "artistmef",
"usernameUnclaimed": "noonewouldeverusethis7",
"checkType": "message",
"alexaRank": 9936,
"tags": [
"photo",
" stock"
]
},
"lightstalking.com": {
"urlMain": "https://www.lightstalking.com",
"engine": "Wordpress/Author",
"usernameClaimed": "jasonrow",
"usernameUnclaimed": "noonewouldeverusethis7",
"alexaRank": 83900,
"tags": [
"blog",
" photo"
]
},
"morguefile.com": {
"absenceStrs": [
"free photographs for commercial use"
],
"presenseStrs": [
"sortName",
" profile-data"
],
"url": "https://morguefile.com/creative/{username}",
"urlMain": "https://morguefile.com",
"usernameClaimed": "thesuccess",
"usernameUnclaimed": "noonewouldeverusethis7",
"checkType": "message",
"tags": [
"photo"
]
},
"wls.social": {
"urlSubpath": "/wls",
"urlMain": "https://wls.social",
"engine": "Wordpress/Author",
"usernameClaimed": "nathaliemariel",
"usernameUnclaimed": "noonewouldeverusethis7",
"tags": [
"blog"
]
},
"steemit": {
"absenceStrs": [
"NotFound__menu"
],
"presenseStrs": [
"profile",
" username"
],
"url": "https://steemit.com/@{username}",
"urlMain": "https://steemit.com",
"usernameClaimed": "apiprincz",
"usernameUnclaimed": "noonewouldeverusethis7",
"checkType": "message",
"alexaRank": 20554,
"tags": [
"news"
]
}
},
"engines": {
maigret/submit.py
+4
View File
@@ -356,6 +356,10 @@ async def submit_dialog(db, url_exists, cookie_file, logger):
chosen_site.name = input("Change site name if you want: ") or chosen_site.name
chosen_site.tags = input("Site tags: ").split(',')
rank = get_alexa_rank(chosen_site.url_main)
if rank:
print(f'New alexa rank: {rank}')
chosen_site.alexa_rank = rank
logger.debug(chosen_site.json)
site_data = chosen_site.strip_engine_data()